See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/272549578

Computational fluid dynamics simulations of interphase heat transfer in a

bubbling fluidized bed

Article in Korean Journal of Chemical Engineering · July 2014

DOI: 10.1007/s11814-014-0022-6

CITATIONS READS

8 721

5 authors, including:

Musango Lungu Jingyuan Sun

The Copperbelt University Zhejiang University
31 PUBLICATIONS 344 CITATIONS 118 PUBLICATIONS 1,077 CITATIONS

SEE PROFILE SEE PROFILE

Yongrong Yang
Zhejiang University
281 PUBLICATIONS 4,009 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jingyuan Sun on 23 October 2016.

The user has requested enhancement of the downloaded file.

Korean J. Chem. Eng., 31(7), 1148-1161 (2014) pISSN: 0256-1115
DOI: 10.1007/s11814-014-0022-6 eISSN: 1975-7220
INVITED REVIEW PAPER INVITED REVIEW PAPER

Computational fluid dynamics simulations of interphase heat transfer

in a bubbling fluidized bed
Musango Lungu, Jingyuan Sun, Jingdai Wang†, Zichuan Zhu, and Yongrong Yang

State Key Laboratory of Chemical Engineering, Department of Chemical and Biological Engineering,
Zhejiang University, Hangzhou 310027, China
(Received 1 November 2013 • accepted 14 January 2014)

Abstract−Numerical simulations based on the Eulerian-Eulerian approach have been performed in the study of in-
terphase heat transfer in a gas solid fluidized bed. The kinetic theory of granular flow (KTGF) has been used to describe
the solid phase rheology. An assessment of drag models in the prediction of heat transfer coefficients shows that no
major difference is observed in the choice of the drag model used. Fluctuations of the interphase heat transfer coefficient
have been found to be closely related to the bubble motion in the bed. Effects of the wall boundary condition, inlet
gas velocity, initial bed height and particle size on the predicted heat transfer coefficient have also been investigated.
Typical temperature profiles in the bed show that thermal saturation is attained instantaneously close to the gas distributor.
Simulated results of the coefficients are in fair agreement with those reported in literature.
Keywords: CFD, Drag Model, Fluidization, Heat Transfer, Simulation

INTRODUCTION
Fluidized bed technology has found popular use in many indus-
trial applications such as gas phase polymerization, combustion,
drying operations and fluid catalytic cracking, to mention a few due
to excellent solids mixing and heat transfer characteristics. These
ideal characteristics are brought about by the vigorous particle motion
and intimate contact of the gas and solid phases. In gas solid fluid-
ized beds, three modes of heat transfer arise: surface-to-bed, inter-
particle and particle-to-gas. The majority of studies reported in litera-
ture have focused on the surface-to-bed heat transfer.
Mickley and Faribanks [1] proposed a packet or cluster renewal
mechanistic model to explain the mechanism of heat transfer from
the bed to the surface. Several other attempts have been made by
various investigators [2,3] in the past to model heat transfer in fluid-
ized beds using mechanistic and empirical models. However, mech-
anistic and empirical models have their shortcomings that render
them unattractive. Mechanistic models are constrained by the as-
sumptions on which they are based, while empirical models work
well only within the range of experimental data on which the model
was arrived at [4]. Furthermore empirical correlations do not help
much in the understanding of the fundamental transport mechanisms
[5].
The advent of high speed computers has given rise to the use of
computational fluid dynamics (CFD) as an indispensable tool in
the study of complex hydrodynamics and transport processes in mul-
tiphase flow systems. Based on the continuum approach, Syamlal
and Gidaspow [6] used the K-FIX computer code to predict the wall
to bed heat transfer coefficients in a bubbling fluidized bed. Their
work revealed that it is possible to predict the large heat transfer
†
To whom correspondence should be addressed.
E-mail: wangjd@zju.edu.cn
Copyright by The Korean Institute of Chemical Engineers.
1148
coefficients without the use of any turbulence model. Several other
studies [4,5,7,8] have been conducted to model wall to bed heat
transfer in bubbling fluidized beds.
Particle-to-particle heat transfer occurs mainly due to conduction
through the contact points between the particles and due to heat ex-
change through the gas layer separating the particles. Delvosalle and
Vanderschuren [9] developed an inter-particle heat transfer model
due to conduction through the gas layer between a hot and a cold
particle. Chang et al. [10] modeled direct particle to particle heat
transfer between different particle classes in a gas solid fluidized
bed, combining the stochastic collision frequency and the direct heat
conduction due to elastic deformation during impact.
Hitherto, very few detailed modeling studies have been dedicated
to particle-to-gas heat transfer in fluidized bed reactors despite the
fact that some industrial reactors like gas phase polymerization reac-
tors are operated adiabatically [11] with the heat of polymerization
being removed as sensible heat by the recycled gas stream. In such
instances no immersed heat transfer tubes are available and so the
wall to bed heat transfer in such cases is not significant. Inaccura-
cies in the measurement of particle and gas temperatures have made
experimental determination of the particle-to-gas heat transfer coef-
ficient difficult. Furthermore, due to the complex flow pattern in
fluidized beds, the particle-to-gas heat coefficients reported in liter-
ature vary a great deal depending on the flow assumption used [12].
The amount of heat exchange between the two phases is unknown
and difficult to measure or guess; thus, computational fluid dynam-
ics can provide an insight into the heat transfer process in fluidized
beds.
Kaneko et al. [13] used a discrete element method (DEM), incor-
porating the reaction and kinetic balances to analyze temperature
behaviors of particles and gas in a fluidized bed reactor for poly-
olefin production. Their simulations revealed hot spot formation
on the distributor near the wall of the fluidizing column. Despite
the increase in computational capabilities, the current available com-
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed 1149

putation capacity renders the DEM approach less attractive in mod-
eling dense fluidized beds such as bubbling fluidized beds. CFD
simulations were carried out by Behjat et al. [14] to investigate hy-
drodynamic and heat transfer phenomena of a bimodal particle mix-
ture in a gas solid fluidized bed. They assumed that solid heat conduc-
tivity includes both direct conduction through the fractional contact
area and indirect conduction through a wedge of gas that is trapped
between the particles. Hamzehei et al. [15] used an Eulerian-Eule-
rian model, incorporating the standard k-ε model turbulence model
in the simulation of unsteady flow and heat transfer in a gas solid
fluidized bed in which hot air entered a cold bed. Their simulation
results were in close agreement with experimental data. Chen et al.
[16] developed a three-dimensional computational fluid dynamics-
population balance method (CFD-PBM) coupled model using an
Eulerian-Eulerian two-fluid model to describe the gas-solid two phase
flow in fluidized bed polymerization reactors. Their results show
that the inlet gas velocity is an important parameter in controlling
reactor temperature fields. As mentioned, 3-D simulations are com-
putationally intensive for parametric studies. Accurate predictions
of heat transfer coefficients are pertinent to the successful design,
scale-up and operation of fluidized beds.
In the present work numerical simulations to predict heat trans-
fer coefficients in a hot gas solid bubbling fluidized bed reactor con-
scribed have also been given in Appendix A. The Gunn correlation
[25] recommended for granular flows has been used to close the
interphase heat exchange between the two phases since it’s appli-
cable for the porosity range of 0.35-1 and a Reynolds number up
to 105. It is well suited for fluidized bed heat transfer modeling. The
next subsection reviews some available Nusselt number correlations.
1. Nusselt Number Correlations
There are numerous empirical correlations reported in literature
for the estimation of both packed bed and fluidized bed fluid-to-
particle heat-transfer coefficients. These correlations relate the Nus-
selt number to the Reynolds number, Prandtl number and in some
cases the voidage. Table 1 below lists some of the correlations. The
correlations have been plotted in Fig. 1 assuming a constant poros-
ity of 0.52 and a constant Prandtl number of 0.7155, and from the
figure two distinct groups can be seen. For the first group, which
includes correlations due to Gunn [25], Ranz and Marshall [26] and
Wakao et al. [27], the Nusselt number approaches a constant value
as the Reynolds number approaches zero. For the second group con-
Table 1. Nusselt number correlations
Ranz and Marshall [26]
Fixed bed Nu=2+1.8Re1/2 s Pr
1/3

taining linear low density polyethylene (LLDPE) particles have been BFB Nu=2+0.6Res Pr1/3 1/2

performed using a verified heat transfer model [5] based on the two- Nelson and Galloway [29]
fluid approach. This approach treats the gas and solid phases as inter- Fixed bed, BFB ⎧ 2ζ 2(1− αg)1/3 ⎫
2ζ + ⎨ ---------------------------------- 1/3 2
- − 2 ⎬tanh ζ
penetrating continua. Though the two fluid model (TFM) is derived ⎩ [ 1 − ( 1− α g ) ] ⎭
from conservation equations of mass, momentum and energy, the Nu = ---------------------------------------------------------------------------
ζ
interphase momentum and heat exchange coefficients are empiri- ----------------------------- − tanh ζ
1 − (1− αg)1/3
cal and hence need detailed investigation. To this effect an investi- where
gation of different drag models has been carried out to study their
1 ε
effect on the predicted coefficients. Influences of the inlet gas veloc- - −1 ---Re1/2
ζ = --------------------
1/3 s Pr
1/3

(1− αg) 2
ity, particle size and initial height on the predicted heat transfer co-
efficient have also been investigated. A typical temperature profile Cybulski et al. [28] Nu=0.07Res
of the gas and solid phase in the reactor has been presented. Fixed bed
Gunn [25] Nu=(7− 10αg +5αg2)(1+0.7Re0.2 s Pr )
1/3

CFD MODEL Nu=+(1.33− 2.4αg +1.2αg )Res Pr1/3 2 0.7

Fixed bed, BFB

The relevant governing equations and the respective constitutive
relations are given in appendix A. The viscous dissipation and work
terms are small when compared to the other terms in the heat balance
(accumulation, conduction, convection and interphase transfer) and
have thus been omitted [7]. The interphase momentum exchange
has been closed using the Syamlal O’Brien [17], Adjusted Syamlal
O’Brien [18], Gidaspow [19], Cao-Ahmadi [20] and Hill-Koch-
Ladd drag [21] models respectively which are given in Appendix
A. The solid phase rheology has been described using the popular
and well known kinetic theory of granular flow [19]. The bulk solid
viscosity, which describes the resistance of the particle suspension
to compression, has been modeled according to Lun et al. [22]. The
shear viscosity, which is composed of three components, kinetic
[19], collision [19] and friction [23], has been considered in the model.
Solids pressure, which represents the normal solid phase forces due
to particle-particle interactions, has been handled using the expres-
sion by Lun et al. [22], meanwhile the radial distribution function,
which describes the probability of particles colliding, has been mod-
eled according to Ma and Ahmadi [24]. The expressions just de- Fig. 1. Nusselt number correlations from literature.
Korean J. Chem. Eng.(Vol. 31, No. 7)
1150 M. Lungu et al.

Table 2. Simulation model parameters in Fluent 6.3

Description Taghipour et al. [33] Present work
3
Particle density 2600 kg/m (glass beads) 918 kg/m3 (LLDPE)
Gas density 1.225 kg/m3 (air) 1.2 kg/m3 (air)
Mean particle diameter 275 μm 348 μm
Particle heat capacity 737 J/kg·K 2600 J/kg·K
Gas heat capacity 1006.43 J/kg·K 994 J/kg·K
Particle thermal conductivity 1.0 W/m·K 0.29 W/m·K
Gas thermal conductivity 0.0242 W/m·K 0.00257 W/m·K
Initial particle temperature 300 K 313 K
Initial gas temperature 300 K 353 K
Static bed height 0.4 m 0.3 m
Bed porosity 0.4 0.52
Bed width 0.28 m 0.15 m
Bed height 1.0 m 1.0 m
Superficial gas velocity 0.46 m/s (7Umf) 0.22-0.66 m/s ( 2Umf −6Umf)

sisting of correlations due to Cybulski et al. [28] and Nelson and was used to validate the model used in the present work. They con-
Galloway [29], the Nusselt number approaches zero as Reynolds ducted experiments in a pseudo 2D using air and glass beads of 250-
number approaches zero, and this is attributed to the negligence of 300 mm diameter (classified as Geldart B). Similarly, LLDPE parti-
axial dispersion [25]. However, for an isolated sphere in an infinite
stagnant fluid the theoretical limiting value of the Nusselt number
is 2 [30]; thus the second group are not suitable as they predict Nu
numbers less than the theoretical minimum with decreasing Re num-
ber. From the first group to which the Gunn correlation belongs we
observe that correlations due to Ranz and Marshall and Wakao et
al. approach the limiting value of 2 well before the Gunn correla-
tion does, implying that the Gunn correlation can be used for a wider
range of Re. The Gunn correlation is applicable for a porosity range
of 0.35-1 and for Re up to 105 as was aforementioned in section 2
of the manuscript, making it ideal for simulations involving heat
transfer in fluidized beds. In addition, the Gunn correlation satisfies
several other less evident asymptotic conditions [5]. Recently, Kuiper’s
research group in the Netherlands validated the Gunn correlation
using DNS simulations [31]. The agreement between the fitted heat
transfer coefficients and the results obtained from the empirical cor-
relation of Gunn were quite reasonable, especially at the higher Rey-
nolds numbers.

SIMULATION SETUP

The geometry used in the simulations is a 2D fluidized bed which

is the same as that used by Sun et al. [32]. The bed has a diameter
of 15 cm and height of 100 cm. The grid is created in GAMBIT
2.3.16 and exported into FLUENT 6.3.26, a CFD commercial soft-
ware package used to execute the numerical simulations. The grid
is a uniform quadratic mesh with 5 mm size interval in both the radial
and axial directions, thus giving 6000 computational cells. An initial
sensitivity study revealed that this mesh size is sufficient for an inde-
pendent mesh solution. Linear low density polyethylene (LLDPE)
is used as the granular media and air as the fluidizing medium. The
minimum fluidization of the particles is 0.11 m/s, while the poros-
ity of the bed at minimum fluidization is 0.52. The properties of
the gas and solid phase are given in Table 2 below.
Experimental voidage data from the work of Taghipour et al. [33] Fig. 2. Geometry of simulation system with boundary conditions.
July, 2014
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed
cles are classified as Geldart B, and thus the fluidization behavior
in the system reported in [33] is supposed to closely resemble that
being simulated. Detailed simulation parameters from Taghipour et
al. have also been listed in Table 2 below.
1. Boundary and Initial Conditions
The initial condition of minimum fluidization is used for all the
simulations. At the bottom of the bed a uniform velocity inlet and
temperature is specified for the gas phase only, as no solids enter
the computational domain. At the top of the column a pressure outlet
boundary condition is assigned. The left and right walls are desig-
nated as no slip for air, and the partial slip wall boundary condition
of Johnson and Jackson [34] has been adopted for the particulate
phase with a specularity coefficient of 0.5, which is commonly used
in simulating bubbling fluidized beds [35]. In addition, the thermal
energy boundary condition at the walls is set to adiabatic. Initial
values of the pressure profile, solid volume fraction and tempera-
tures are patched in the flow field. Fig. 2 below gives a summary
of the geometry, initial and boundary conditions used.
2. Simulation Procedure and Method
The simulations were run for 20 seconds real time in the unsteady
state mode and quantities were time averaged for the last 15 seconds
as it took 5 seconds for the initial start-up effects to subside. Sec-
ond-order spatial discretization was used for momentum, granular
temperature and energy, second-order implicit for the transient for-
mulation and the QUICK scheme for volume fraction. For pres-
sure velocity coupling the semi implicit method for pressure linked
equations (SIMPLE) scheme was used. A fixed time step of 0.0005 s
was used throughout the simulations. The default convergence crite-
ria of 10−3 were used for all quantities except for the energy equations,
which were set to 10−6.
RESULTS AND DISCUSSION
1. Mesh Refinement Study
Investigation of the effect of the mesh size on the solution results
is one of the first steps performed in CFD studies. Ideally a mesh
independent solution is one which does not change with further mesh
refinement. Grid sensitivity was carried out with interval spacing
of 2.5 mm, 5 mm and 10 mm, respectively, in both the radial and
Fig. 3. Time averaged interphase heat transfer coefficient for dif-
ferent grid resolution (Ug=0.44 m/s, Hill et al. drag model).
1151
Table 3. Average interphase heat transfer coefficient for different
number of grids
Number of grids Predicted heat transfer coefficient (W/m3 ·K)
24000 663624.44
06000 664145.58
01500 659973.79
Fig. 4. Contours of the interphase heat transfer coefficient for dif-
ferent mesh resolutions (Ug =0.44 m/s, t=20 seconds).
axial direction corresponding to 24000, 6000 and 1500 grids, respec-
tively. Fig. 3 shows the time averaged (5-20 sec) and spatially aver-
aged (along the width of the reactor) volumetric interphase heat trans-
fer coefficient profile along the height of the reactor obtained using
the three mesh sizes. Qualitatively, all three mesh sizes give a similar
trend. Calculated average values of the coefficients show that grids
of size 24000 and 6000 give nearly the same value, indicating that a
mesh size of 5 mm gives a grid independent solution. Table 3 shows
the calculated average values of the transfer coefficient in the bed.
Fig. 4 gives the instantaneous heat transfer coefficient in the bed
after 20 seconds and at a velocity of 0.44 m/s. The figure further
confirms that meshes with 24000 and 6000 grids show similar char-
acteristics in the dense bed. Consequently, a mesh with 6000 grids
was adopted and used in the remainder of the studies.
2. Drag Model Study
2-1. Qualitative Comparison of Drag Models
The observed high heat transfer rates in fluidized bed reactors
are attributed to the hydrodynamics [36]. The drag force plays an
important role in influencing the hydrodynamics in the fluidized
bed and therefore needs investigation. Fig. 5 shows the contour plots
of the solids volume fraction obtained using different drag models
after 20 seconds of simulation and at a superficial gas velocity of
0.44 m/s. Clearly, all the models give a similar qualitative result with
respect to bed expansion and bubble shape. At 20 seconds the bed
is already at steady state and thus the contour plots represent the
typical fluidization behavior in the bed. A comparison between the
adjusted Syamlal O’Brien and Syamlal O’Brien models reveals that
the former predicts a slightly higher bed expansion and smaller aver-
age bubble size compared to the latter. Small bubble size translates
to slower bubble rise velocity, and consequently a higher expanded
Korean J. Chem. Eng.(Vol. 31, No. 7)
1152 M. Lungu et al.
Fig. 5. Contours of the solid volume fraction for different drag mod-
els (Ug =0.44 m/s, t=20 s).
bed height [37]. Generally, smaller bubbles can be observed close
to the plate distributor and large bubbles are visible at higher heights
near the freeboard for all the drag models. This is in agreement with
the experimentally observed trend that bubbles grow as they rise in
the bed due to coalescence.
Although there have been extensive comparisons of drag models
in hydrodynamics [38,39], parallel work for heat transfer modeling
in fluidized beds is scanty [40]. Fig. 5 shows a plot of the time aver-
aged (5-20 sec) and spatially averaged (along the width of the reac-
tor) volumetric interphase heat transfer coefficient profile along the
height of the reactor obtained using different drag models at a superfi-
cial velocity of 0.44 m/s. The heat transfer coefficients were calcu-
lated using a user-defined function incorporated into FLUENT 6.3.26
and the simulation data was exported into Matlab for processing.
Qualitatively, all drag models give a similar trend in the interphase
heat transfer coefficient profile.
All the drag models show that near the inlet of the bed the heat
transfer coefficient rises sharply at first, reaching a characteristic
peak and then varies unsteadily until about a height of 0.45 m, after
which it then starts to decline gradually until it reaches zero. The
Cao Ahmadi drag model predicts the least value of the coefficient
in the dense phase and the largest in the lean phase. Relative to other
models the Cao Ahmadi model predicts a large voidage in the dense
Fig. 7. Plot of (a) local Nu number dependence on voidage for different drag models at a height of 0.3 m above the plate distributor at 20 s
(b) time and spatially averaged Nu number dependence on voidage in the fluidized bed.
July, 2014
Fig. 6. Comparison of interphase heat transfer coefficient profiles
for different drag models (Ug =0.44 m/s).
bed from a plot of spatially and time averaged bed voidage for dif-
ferent drag models (available on demand), which is not presented
in the paper to avoid repetition of data and for the sake of brevity.
A higher voidage means that heat transfer is poor, and thus the ob-
served low heat transfer coefficient profile in the dense bed. In bub-
bling fluidized beds particles are seldom blown out of the reactor
and thus a reduction of particle concentration in the dense phase
leads to a corresponding increase in the upper part of the reactor
and thus the observed high heat transfer coefficient predicted by
the Cao Ahmadi drag model in the lean phase. When the Syamlal
O’Brien model is adjusted to predict the experimentally observed
minimum fluidization velocity, it is observed that the resulting model
(adjusted Syamlal O’Brien) predicts a reduced value of the heat trans-
fer coefficient in the dense bed and a corresponding increase in the
lean phase. As aforementioned, the adjusted Syamlal O’Brien model
predicts an expanded bed height, which results in the reduction of
particle concentration in the dense bed and a corresponding parti-
cle concentration increase in the upper part of the reactor due to the
slow rising bubbles. Since the volumetric interphase heat transfer
coefficient is directly proportional to the solid volume fraction, it
will exhibit a similar behavior to the particle concentration profile.
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed
In addition, small bubble sizes in gas solid fluidized beds seem not to
enhance solids mixing, which results in the observed low heat transfer
coefficient in the dense phase. The Gidaspow, Syamlal O’Brien and
Hill Koch Ladd drag models predict similar profiles of the heat transfer
coefficient along the height of the reactors which are all close to
each other with insignificant differences. In bubbling fluidized beds
the lean phase contains very few solid particles or none at all; there-
fore, the heat transfer coefficient drops to zero, indicating that gas
(air) is a poor transporter of heat as shown by Eq. (A.40).
Fig. 7 displays local and time-averaged plots of the Nu depen-
dence on voidage for different drag models. As expected, the drag
models have very little influence on the predicted profiles. Both
the local and time-averaged plots display similar characteristics of
exponential decay with increasing voidage. At low voidage the Nus-
selt number is high, and consequently higher heat transfer coeffi-
cients are predicted. Typically, low voidage exists in the dense region
of the bubbling fluidized bed, and it is in this region that high heat
transfer rates are experienced. As the voidage approaches one, the
Nusselt number approaches the limiting value of two.
2-2. Quantitative Comparison of Drag Models
The time-averaged voidage predicted by the various drag models
has been compared against experimental data reported by Taghipour
et al. [33] From Fig. 8, the simulated result of the time averaged
voidage profile at a height of 0.2 m above the distributor plate at a
gas velocity of 0.46 m/s compares fairly well with the experimen-
tal data. Both the simulated and experimental results show that the
voidage in the core of the bed is high with a slight decrease near
the wall. This profile correctly demonstrates that preferred path of the
bubbles in the center of the bed. Cao Ahmadi predicts the largest
value of the time-averaged voidage along the whole radial distance
of the bed. The adjusted Syamlal O’Brien gives the second largest
value of the voidage but only on the left plane of the reactor. On
the right plane it gives similar results to other drag models. The Syam-
lal O’Brien model gives good results in comparison to the experi-
mental data near the walls but not in the core of the bed. The Hill
Koch Ladd model predicts consistent and nearly symmetrical values
of the voidage in the core of the bed. Overall, the differences between
Fig. 8. Experimental and Simulated time averaged voidage pro-
file at a height of 0.2 m above the distributor and superfi-
cial gas velocity of 0.46 m/s.
1153
Fig. 8. Drag force for different drag models as a function of poros-
ity for the conditions listed in Table 2.
the drag models are not very large. Consequently, the Hill Koch
Ladd drag model was adopted and used in the rest of the simula-
tions.
Fig. 8 is a plot of the drag force as a function of the porosity for
different drag models at a relative velocity of 1m/s. The drag force,
which is a function of the porosity, slip velocity and properties of
the gas and solid phase, has been calculated using the conditions
listed in Table 1. For bubbling fluidized beds the mean voidage range
is 0.45 to 0.65 [41]; however, in our study we investigated the be-
havior of the drag correlations over a wider range from a minimum
porosity of 0.52 representing packed bed to a maximum porosity of
0.95. From the plot it is clear that the shape of the drag force pro-
file predicted by the Cao Ahmadi correlation differs from the rest.
The shape of the curve resembles the letter S rotated 90o anticlock-
wise. At the minimum fluidization condition, the Cao Ahmadi model
predicts the largest drag force and with increasing porosity it de-
creases and crosses the Adjusted Syamlal O’Brien at a porosity of
0.56 and the other drag models between the porosity ranges of 0.7
to 0.82 after which it gives the least value. The Gidaspow, Syamlal
O’Brien and the Hill Koch Ladd drag models show a similar be-
havior throughout the whole porosity range and give values close
to each other with the drag force reducing with increasing bed void-
age. A closer look at the Gidaspow model reveals a ‘switch’ inherent
in the drag model at a porosity of 0.8. The drag force predicted by
the three drag models is significantly lower than that predicted by
the adjusted Syamlal O’Brien drag model. This is in agreement with
Vejahati et al. [39] who observed that adjustment of drag models
based on the minimum fluidization velocity results in prediction of
higher values of drag coefficient and consequently drag force through-
out the whole porosity range. In conclusion, the drag model study
shows that very minor differences are observed with the choice of the
drag model used for simulating heat transfer in the gas solid bubbling
fluidized bed.
The average bed volume fraction in bubbling beds can be conve-
niently obtained from the relationship:
dp
− ------ = ρs( 1− αg)g (1)
dy
From which in turn the average specific area 6αs, ave/ds can be esti-
Korean J. Chem. Eng.(Vol. 31, No. 7)
1154 M. Lungu et al.
Fig. 9. (a) Heat transfer coefficient profile for different specularity coefficients, (b) heat transfer coefficient for different particle-wall
restitution coefficient.
mated assuming the particles are spherical. From our simulation
the average specific area was estimated to be 7,049 m2/m3 at a superfi-
cial gas velocity of 0.44 m/s using the Hill Koch Ladd drag model.
High heat transfer rates in fluidized beds are mainly due to the large
interfacial area, typically in the range of 3,000-45,000 m2/m3 as op-
posed to the low particle-gas heat transfer coefficients, which are
typically in the range of 6-23 W/m2K [42]. Our simulation pre-
dicted an average interphase heat transfer coefficient of 664 145 W/
m3K at a velocity of 0.44 m/s. From this value, the predicted parti-
cle-to-gas heat transfer coefficient is only 94.22 W/m2K.
3. Effect of Wall Boundary Condition on the Predicted Heat
Transfer Coefficient Profile
As has been mentioned, the wall boundary condition is taken care
of using the Johnson and Jackson semi empirical equation. To inves-
tigate this effect, two parameters were varied: the specularity coeffi-
cient, φp and the particle wall restitution coefficient, ew. The specularity
coefficient is a measure of fraction of collisions which transfer mo-
mentum to the wall and assumes values between zero and one with
zero indicating perfectly specular collisions and one perfectly dif-
fuse collision. In our study φp values of 0, 0.5 and 1 representing
free slip, partial slip and no slip conditions were tried. Meanwhile,
ew characterizes the dissipation of solid kinetic energy by collisions
with the walls. Values of 0.2, 0.55 and 0.9 were used to investigate
ew. Fig. 9(a) shows the time averaged (5-20 sec) and spatially aver-
aged (along the width of the reactor) volumetric interphase heat trans-
fer coefficient profile along the height of the reactor obtained using
the three slip conditions. Clearly, the free slip condition is different
from the partial slip and no slip conditions, which result in similar
profiles of the transfer coefficient. The free slip condition entails
that the contact between the solid and the wall is frictionless, leading
to higher downward solid concentration and higher velocity close
to the wall [35]. This leads to accumulation to solids at the bottom
of the reactor, and consequently a higher heat transfer coefficient is
observed. The increase of solid particles in the bottom of the reactor
is accompanied by a corresponding decrease of particle concentra-
tion in the upper part of the reactor and hence the observed reduced
heat transfer coefficient in the upper part of the reactor. The absence
of friction in the modeling studies leads to a different solids circu-
lation pattern [43], which in turn affects the heat transfer coefficient
profile in the bed. The partial slip and free slip conditions give nearly
identical profiles of the heat transfer coefficient. Inclusion of fric-
July, 2014
tion leads to a reduction of the solids concentration and velocity
near the wall since particles experience resistance to movement.
This action leads to less accumulation of solid particles at the bot-
tom of the reactor and therefore a lower heat transfer coefficient.
At the upper part an increase of the heat transfer coefficient is ob-
served that obviously is as a result of an increase in the particle con-
centration. Therefore, to predict accurate values of the heat transfer
coefficient, care must be taken in the specification of wall bound-
ary conditions.
Fig. 9(b) shows the time and spatially averaged heat transfer co-
efficient at a velocity of 0.44 m/s and ew values of 0.2, 0.55 and 0.9.
No apparent differences in the profiles are observed for the three ew
values, meaning that the particle wall coefficient of restitution has
no effect on the model results. Similarly, Li et al. [35] did not observe
any appreciable difference in the hydrodynamic behavior using dif-
ferent ew values in their study. Consequently, a specularity coeffi-
cient of 0.5 and a particle wall coefficient restitution of 0.2 were
used throughout our simulations.
4. Heat Transfer Coefficient Fluctuations
The transient heat transfer coefficient exhibits oscillatory behav-
ior similar to the commonly encountered pressure fluctuations in
fluidized beds owing to the bubble motion in the bed. “Numerical
probes” were positioned at heights of 0.15 m, 0.30 m, and 0.45 m
above the distributor to follow the time evolution of the heat trans-
fer coefficient. As the rising bubbles approach the probe, the inter-
phase heat exchange coefficient drops sharply due to low thermal
conductivity and heat capacity of the gas phase. As the bubbles fur-
ther rise, the interphase heat transfer coefficient sharply rises behind
the bubbles due to the increased particle concentration in the wake.
This mechanism continues, thus giving rise to the oscillatory behav-
ior. Experimental studies with single bubbles rising in a gas-solid
fluidized bed have shown that the bubble wake plays an important
role in particle circulation and heat transfer [44]. The same mecha-
nism is at play in gas solid fluidized beds with multiple bubbles.
Using spectral analysis, the heat transfer coefficient fluctuations can
be analyzed to obtain the dominant frequency. The dominant fre-
quency is the highest power intensity in the PSD and is found to
correspond to different mechanisms or hydrodynamic behavior vari-
ation including bubbles formation and eruption [45]. Fig. 10 shows
the local instantaneous heat transfer coefficient fluctuations and the
corresponding power spectral density (PSD) at heights of 0.15 m,
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed 1155

Fig. 10. Time series of the heat transfer coefficient fluctuations and corresponding power spectral density at a height of (a) 0.15 m (b)
0.30 m and (c) 0.45 m above the distributor.

0.30 m and 0.45 m above the distributor. The PSD is obtained di-
rectly from the heat transfer coefficient data using the fast Fourier
transform (FFT) technique. From the figure a broad band of peak
frequencies can be observed between 0 and 2.5 Hz with a domi-
nant frequency of about 0.9 Hz. At a height of 0.15 m, the domi-
nant frequency is not very visible as the position is closest to the
distributor where the bubbles formation occurs, but with an increase
in the distance from distributor the dominant frequency becomes
distinct. Baskakov et al. [46] derived a formula for the natural fre-
quency in a bubbling fluidized bed, which is the frequency associ-
ated with a single mechanism such as particle vibration or sponta-
neous bed oscillation. The formula is given as f=(1/π) g/Hmf , where
g is the acceleration due to gravity and Hmf is the bed height at min-
imum fluidization. Plugging in the variables yields a natural fre-
quency of 1.82 Hz, which is larger than the dominant frequency. Fig. 11. Effect of the superficial gas velocity on the predicted heat
The dominant frequency of the local heat transfer coefficients is transfer coefficient.
Korean J. Chem. Eng.(Vol. 31, No. 7)
1156 M. Lungu et al.

lower than the natural frequency. Therefore, a single oscillation mech-

anism cannot be responsible for the local fluctuations, which fur-
ther proves that the heat transfer coefficient fluctuations are associated
with the bubble motion in the bed.
5. Effect of Superficial Gas Velocity
Fig. 11 shows the time-averaged profile of the volumetric inter-
phase heat transfer coefficient along the height of the reactor obtained
at different superficial gas velocities. Velocities of 0.22 m/s, 0.44 m/
s and 0.66 m/s which correspond to 2Umf, 4Umf and 6Umf have been
used in the present study. An increase in velocity results in the re-
duction of the particle concentration in the lower part of the reactor
and a corresponding increase in the upper part. From the Gunn cor-
relation we can deduce that an increase in the void fraction leads to
a reduction in the Nusselt number, and consequently the heat trans-
fer coefficient and vice versa. As the velocity is increased from 2Umf,
4Umf and 6Umf, the average heat transfer coefficient increases from
659 288.66 W/m3·K, 664 145.58 W/m3·K and 667 407.63 W/m3·K
respectively. Though a higher superficial velocity results in increased
bed voidage, it also leads to higher particle Reynolds number, which
enhances heat transfer in the bed. The difference in the three values
is quite small due to the opposing effects of the bed voidage and
the high particle Reynolds number at higher velocities.
6. Effect of the Particle Size
Fig. 12 illustrates the effect of the particle size on the predicted
heat transfer coefficient at a constant superficial gas velocity of 0.44
m/s. Two particle sizes of 223 µm and 631 µm have been used in
addition to the default particle size of 348 µm. An increase in the
particle size results in the reduction of the heat transfer coefficient
as well as the bed height. A small particle size leads to an increase
in the drag force, which explains the increase in the bed height with
a decrease in the particle size. The expanded bed height influences
the gas residence time and ultimately the predicted heat transfer co-
efficient. Fig. 13 shows the drag force as a function of the voidage
for the three particle sizes at a relative velocity of 1 m/s. The dif-
ference in the drag force experienced by the three particle sizes is
more pronounced at lower void fractions that are in the dense bed,
and this is expected since the interstitial velocity is high due to the
smaller voidage and the particle movement is restricted. Note that
the drag force is based on a unit volume. The drag force is reduced
with increasing porosity as the particles become more mobile. As
Fig. 13. Drag Force for different particle sizes.
[47,48] have demonstrated, a small particle size creates larger bub-
bles. Wang et al. [49] using acoustic emission (AE) investigated
the influence of particle size on the flow structure in fluidized beds,
and they concluded that fluidization with small particles leads to
more gas bubbles being formed at the distributor, causing a larger
bed expansion.
7. Effect of Initial Bed Height
The effect of the initial bed height was investigated using bed
heights of 0.15 m, 0.30 m and 0.45 m respectively. Table 3 shows
the values predicted for average heat transfer coefficient, height of
expanded bed, expansion bed ratio, and average bed solid volume
fraction and gas residence time at the three initial bed heights.
The gas residence time can be conveniently calculated from the
expression [50]:
Hexp
tres = --------------------
- (2)
Ug/αg, ave
From the table we notice that a decrease in the initial bed height
results in an increase in the bed expansion ratio and a decrease in
the gas residence time. In bubbling fluidized beds, the bubble rise
velocity increases with an increase in the bubble diameter based on
the simple two phase theory of Davidson and Harrison [12] as the
bubbles move up the bed. Thus, a reactor with a small initial static
bed height will have a smaller average bubble size and a high bed
expansion ratio [50] as a result of the slow bubble rise velocity. A
high bed expansion ratio is synonymous with an average high void
fraction and thus a reduced volumetric interphase heat transfer co-
efficient. Thus, to promote heat transfer in a fluidized bed, a com-
pact bed accompanied by a large gas residence time is desirable.

Table 4. Effect of initial bed height on the predicted heat transfer

coefficient
Hmf hsg hexp δ =(Hexp − Hmf)/Hmf αg, ave tres
(m) (W/m3K) (m) (-) (-) (s)
0.45 1 007 000 0.74 0.64 0.62 1.04
0.30 672 185.19 0.51 0.7 0.64 0.74
Fig. 12. Effect of particle size on predicted heat transfer coefficient. 0.15 337 393.47 0.28 0.87 0.69 0.44

July, 2014
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed 1157

Fig. 14. (a) Temperature profile in Gas phase, (b) temperature profile in solid phase.

8. Temperature Profile of Gas and Solid Phase the interphase heat transfer coefficient are closely linked to the bubble
In the present work the gas enters the bed at a temperature of 313 motion in the bed. With an increase in the initial bed height, an in-
K and the solids are initially at 353 K, representing typical temper- crease in the heat transfer coefficient is observed. An increase in
atures in a polymerization reactor [51]. Fig. 14(a) and Fig. 14(b) the superficial gas velocity leads to an appreciable increase in the
show the temperature profiles of the gas phase and solid phase after heat transfer coefficient. With the reduction in the particle size, an
20 seconds of simulations and at a superficial gas velocity of 0.44 increase in the heat transfer coefficient is observed together with an
m/s. Fig. 13(a) shows that the gas is rapidly heated to about 351 K, increase in the expansion of the bed. The temperature profiles of the
showing that the temperature of the gas follows the temperature of gas phase and solid phase shows that thermal equilibrium is reached
the particle. Thus, the temperature gradient is greatest at the inlet to instantaneously close to the gas distributor. Thus our study has dem-
the bed due to the large temperature driving force, i.e., the temper- onstrated that CFD is a powerful tool capable of capturing the com-
ature difference between the two phases. plex heat transfer process in gas solid fluidized beds and is a useful
Van Heerden et al. [52] proposed an equation to calculate the ther- tool in the design and scale up of fluidized bed reactors.
mal relaxation time, τg, in a fluidized bed which is the time required
to establish a stationary gradient between the solid and gas phase ACKNOWLEDGEMENTS
and is given by:
The authors sincerely wish to thank the financial support pro-
2
2.3ρgcpgds vided by The Project of National Natural Science Foundation of
τg = ------------------------
- (3)
36κg, 0 China (21236007) and National Basic Research Program of China
(2012CB720500).
Substitution of data into the above equation yields a thermal relax-
ation time of 359 µs; multiplying this value with a velocity of 0.44 NOMENCLATURE
m/s gives a distance of 0.0158 cm. Thus, a stationary gradient is
established instantaneously at the bottom of the bed very close to CD : particle drag force coefficient [-]
the distributor; meanwhile simulations as shown by the temperature cp : specific heat capacity at constant pressure [J/kg·K]
profile of the gas phase show that a stationary gradient is estab- ds : mean particle diameter [m]
lished within 0.1 m. g : acceleration due to gravity [m/s2]
h : fluid-particle heat transfer coefficient [W/m2 ·K]
CONCLUSIONS hsg : interphase heat transfer coefficient [W/m3 ·K]
H : enthalpy [J/kg]
Simulations of interphase heat transfer in a gas solid fluidized Nu : Nusselt number [-]
bed have been performed using the Euler-Euler approach. An as- p : pressure [pa]
sessment of drag models shows that all drag models show a similar Pr : Prandtl number
trend in the interphase heat transfer profile in the bed. Thus we can Res : particle Reynolds number [-]
conclude that the drag models used do not have a significant effect t : time [s]
on the predicted transfer coefficients. The large heat transfer rates tres : gas residence time [s]
in the bed are strongly influenced by the large specific interfacial T : temperature [K]
areas rather than the particle-to-gas heat transfer coefficient, which Ug, u : superficial gas velocity [m/s]
is low. Care must be taken in specification of wall boundary condi- us, slip : slip velocity of particle at the wall [m/s]
tions, especially the specularity coefficient, which significantly influ- us, w : tangential velocity of the particle at the wall [m/s]
ences the heat transfer coefficient profile in the bed. Fluctuations of y : vertical coordinate [m]
Korean J. Chem. Eng.(Vol. 31, No. 7)
1158 M. Lungu et al.

Greek Letters kinetic theory descriptions, Academic Press (1994).

α : volume fraction [-] 20. J. Cao and G. Ahmadi, Int. J. Multiphase Flow, 21, 1203 (1995).
αs, max : solid volume fraction at maximum packing [-] 21. S. Benyahia, M. Syamlal and T. J. O'Brien, Powder Technol., 162,
βgs : gas/solid momentum exchange coefficient [kg/m2s] 166 (2006).
ξs : bulk viscosity [kg/m·s] 22. C. Lun, S. Savage, D. Jeffrey and N. Chepurniy, J. Fluid Mech., 140,
δ : bed expansion ratio [-] 223 (1984).
Θs : granular temperature [m2/s2] 23. D. G. Schaeffer, Journal of Differential Equations, 66, 19 (1987).
Θw : granular temperature at the wall [m2/s2] 24. D. Ma and G. Ahmadi, J. Chem. Phys., 84, 3449 (1986).
φp : specularity coefficient [-] 25. D. Gunn, Int. J. Heat Mass Transfer, 21, 467 (1978).
ρ : density [kg/m3] 26. W. Ranz and W. Marshall, Chem. Eng. Prog., 48, 141 (1952).
=
τk : stress tensor of phase k [pa] 27. N. Wakao, S. Kaguei and T. Funazkri, Chem. Eng. Sci., 34, 325
τg : thermal relaxation time [s] (1979).
κ : thermal conductivity [W/m·k] 28. A. Cybulski, M. J. Van Dalen, J. W. Verkerk and P. J. Van Den Berg,
γs : collisional dissipation of solid fluctuating energy [kg/m·s3] Chem. Eng. Sci., 30, 1015 (1975).
γw : collisional dissipation of solid fluctuating energy at the wall 29. P. A. Nelson and T. R. Galloway, Chem. Eng. Sci., 30, 1 (1975).
[kg/m·s3] 30. B. Bird, W. Stewart and E. Lightfoot, Transport phenomena, revised
2nd Ed., John Wiley & Sons, Inc. (2006).
Subscripts and Superscripts 31. N. G. Deen, S. H. L. Kriebitzsch, M. A. Van der Hoef and J. A. M.
g : gas Kuipers, Chem. Eng. Sci., 81, 329 (2012).
k : phase k, solid or gas 32. J. Sun, Y. Zhou, C. Ren, J. Wang and Y. Yang, Chem. Eng. Sci.,
s : solid phase 66, 4972 (2011).
33. F. Taghipour, N. Ellis and C. Wong, Chem. Eng. Sci., 60, 6857 (2005).
REFERENCES 34. P. Johnson and R. Jackson, J. Fluid Mech., 176, 67 (1987).
35. T. Li, J. Grace and X. Bi, Powder Technol., 203, 447 (2010).
1. H. S. Mickley and D. F. Fairbanks, AIChE J., 1, 374 (1955). 36. R. Yusuf, M. C. Melaaen and V. Mathiesen, Chem. Eng. Technol.,
2. R. S. Brodkey, D. S. Kim and W. Sidner, Int. J. Heat Mass Trans- 28, 13 (2005).
fer, 34, 2327 (1991). 37. S. Cloete, S. T. Johansen and S. Amini, Powder Technol., 239, 21
3. R. S. Figliola and D. E. Beasley, Chem. Eng. Sci., 48, 2901 (1993). (2013).
4. R. Yusuf, B. Halvorsen and M. C. Melaaen, Int. J. Multiphase Flow, 38. C. Loha, H. Chattopadhyay and P. K. Chatterjee, Chem. Eng. Sci.,
42, 9 (2012). 75, 400 (2012).
5. J. Kuipers, W. Prins and W. Swaaij, AIChE J., 38, 1079 (1992). 39. F. Vejahati, N. Mahinpey, N. Ellis and M. B. Nikoo, Can. J. Chem.
6. M. Syamlal and D. Gidaspow, AIChE J., 31, 127 (1985). Eng., 87, 19 (2009).
7. D. Patil, J. Smit, M. Sint Annaland and J. Kuipers, AIChE J., 52, 58 40. L. M. Armstrong, S. Gu and K. H. Luo, Int. J. Multiphase Flow,
(2006). 36, 916 (2010).
8. L. Armstrong, S. Gu and K. Luo, Int. J. Heat Mass Transfer, 53, 4949 41. J. R. Grace, Powder Technol., 113, 242 (2000).
(2010). 42. J. S. M. Botterill, Fluid-bed heat transfer: Gas-fluidized bed behav-
9. C. Delvosalle and J. Vanderschuren, Chem. Eng. Sci., 40, 769 (1985). iour and its influence on bed thermal properties, Academic Press,
10. J. Chang, G. Wang, J. Gao, K. Zhang, H. Chen and Y. Yang, Pow- London (1975).
der Technol., 217, 50 (2012). 43. C. Loha, H. Chattopadhyay and P. K. Chatterjee, Particuology,
11. Y. Yang, J. Yang, W. Chen and S. Rong, Ind. Eng. Chem. Res., 41, 11, 673 (2013).
2579 (2002). 44. J. Tuot and R. Clift, AIChE Symp. Ser., 78 (1973).
12. D. Kunii and O. Levenspiel, Fluidization engineering, Butterworth- 45. O.-a. Jaiboon, B. Chalermsinsuwan, L. Mekasut and P. Piumsom-
Heinemann Boston (1991). boon, Powder Technol., 233, 215 (2013).
13. Y. Kaneko, T. Shiojima and M. Horio, Chem. Eng. Sci., 54, 5809 46. A. P. Baskakov, V. G. Tuponogov and N. F. Filippovsky, Powder
(1999). Technol., 45, 113 (1986).
14. Y. Behjat, S. Shahhosseini and S. H. Hashemabadi, International 47. O. Olaofe, M. Van der Hoef and J. Kuipers, Chem. Eng. Sci., 66,
Communications in Heat and Mass Transfer, 35, 357 (2008). 2764 (2011).
15. M. Hamzehei, H. Rahimzadeh and G. Ahmadi, Ind. Eng. Chem. 48. X. Z. Chen, D. P. Shi, X. Gao and Z. H. Luo, Powder Technol., 205,
Res., 49, 5110 (2010). 276 (2011).
16. X. Z. Chen, Z. H. Luo, W. C. Yan, Y. H. Lu and I. S. Ng, AIChE J., 49. J. Wang, C. Ren, Y. Yang and L. Hou, Ind. Eng. Chem. Res., 48, 8508
57, 3351 (2011). (2009).
17. M. Syamlal and T. J. O’Brien, AIChE Symposium Series, 85, 22 50. S. Cloete, S. T. Johansen and S. Amini, Powder Technol., 239,
(1989). 21 (2013).
18. M. Syamlal and T. J. O’Brien, The derivation of a drag coefficient 51. M. L. DeChellis, J. R. Griffin and M. E. Muhle, US Patent, 5,405,922
formula from velocity-voidage correlations, US Department of (1995).
Energy, Morgantown (1987). 52. C. Van Heerden, A. Nobel and D. Van Krevelen, Ind. Eng. Chem.,
19. D. Gidaspow, Multiphase flow and fluidization: Continuum and 45, 1237 (1953).
July, 2014
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed 1159
4.14 1.28
53. R. J. Hill, D. L. Koch and A. J. Ladd, J. Fluid Mech., 448, 213 (2001). A = αg and B = Pαg for αg ≤ 0.85 (A.12)
54. R. J. Hill, D. L. Koch and A. J. Ladd, J. Fluid Mech., 448, 243 (2001). 4.14 Q
55. D. Ma and G. Ahmadi, J. Chem. Phys., 84, 3449 (1986). A = αg and B = αg αg for >0.85 (A.13)

P=0.8 and Q=2.65

APPENDIX A TWO FLUID MODEL The adjusted Syamlal O’Brien drag function is obtained by cali-
AND CONSTITUTIVE RELATIONS brating the Syamlal O’Brien drag model such that it predicts the ob-
served experimental minimum fluidization velocity. This is achieved
Mass conservation for phase k (k=g for gas and s for solid phase) by adjusting the parameters P and Q above using a Microsoft Excel
∂- spreadsheet.
--- ( α ρ ) + ∇ ⋅ ( αk ρ k u k ) = 0 (A.1) Gidaspow drag model [19]
∂t k k
3 αsαgρg ug − us- −2.65
Vk = ∫ αkdV (A.2) βgs = --- CD -------------------------------- αg for αg > 0.8 (A.14)
V 4 ds
2
Where αs μg αg ρg u g − us
βgs =150----------
- +1.75---------------------------
2
for αg ≤ 0.8 (A.15)
αg d s ds
n
∑ αk =1 (A.3)
24 0.687
k=1
cD = --------------[1+ 0.15(αgRes ) ], Res < 1000 (A.16)
αgRes
Momentum conservation for solid phase
cD=0.44, Res>1000 (A.17)
∂
---- (αsρsus) + ∇ ⋅ (αsρsusus) = ∇ ⋅ τ=s
∂t Cao Ahmadi drag model [20]
− αs∇P − ∇Ps + βgs(ug − us) + αsρsg (A.4) 0.75
18μgαs [1+ 0.1(Res ) ]
βgs = ----------------
- ------------------------------------------------
2.5α
- (A.18)
ds (1− ( α /α
Momentum equation for gas phase: s s, max ) )
s, max

∂- Hill et al. drag model [53,54]

--- (α ρ u ) + ∇ ⋅ (αgρgugug) = ∇ ⋅ τ=g
∂t g g g Unlike the other drag models which are empirical in nature, the
− αg∇P − βgs(ug − us) + αgρgg (A.5) Hill et al. drag model is based on detailed lattice Boltzmann simula-
tions, thereby making it a kinetic theory based model. In the present
Thermal energy conservation for solid phase
work the modified Hill et al. model [21], which covers the full range
∂- of void fractions and Reynolds numbers encountered in fluidized
--- (α ρ H ) + ∇ ⋅ (αsρsusHs) = − ∇ ⋅ αsκ s∇Ts + hsg(Tg − Ts) (A.6)
∂t s s s beds, has been adopted and is presented below.
2 2
Thermal energy conservation for gas phase βgs =18Fμgαg(1− αg)/ds (A.19)

∂- F2 −1
--- (α ρ H ) + ∇ ⋅ (αgρgugHg ) = − ∇ ⋅ αgκg∇Ts − hsg(Tg − Ts ) (A.7) F =1+ 3/8Res, αs ≤ 0.01 and Res ≤ ---------------------
-
∂t g g g ( 3/8 − F3)
2
Granular Temperature equation 2 F3 + F3 − 4F1(F0 − F2)
F = F0 + F1Res ,αs > 0.01 and Res ≤ ----------------------------------------------------
-
2F1
∂-
3--- --- (A.20)
(α ρ Θ ) + ∇ ⋅ (αsρsusΘs) = (− PsI + τs):
2 ∂t s s s ⎧ (F2 −1)
⎪ αs ≤ 0.01 and Res > ----------------------
∇us − ∇ ⋅ ks∇Θs − γs − 3βgsΘs (A.8) ⎪ (3/8 − F3 )
F = F2 + F3Res⎨
2
⎪ F3 + F3 − 4F1(F0 − F2)
CONSTITUTIVE EQUATIONS FOR INTERPHASE ⎪ αs > 0.01 and Res > ----------------------------------------------------
-
MOMENTUM TRANSFER ⎩ 2F1

⎧ 1+ 3 αs/2 + (135/64 )αsIn(αs ) +17.14αs

Syamlal O’Brien drag model [17] ⎪ (1− w) ----------------------------------------------------------------------------------------------
-
⎪ 1+ 0.681αs − 8.48αs + 8.16α
2 3

3 αsαgρg ⎛ Re ⎪
βgs = --- ---------------- C --------s⎞ u − u (A.9) ⎪ αs
4 v2 d D⎝ vrs ⎠ g s F0 = ⎨ + w 10 ------------------ -3 0.01 < αs < 0.4 (A.21)
rs s
⎪ (1− αs)
2 ρ g ug − us ds ⎪
CD = ⎛0.63 + --------------------⎞ , Res = -------------------------
4.8 - (A.10) ⎪ αs
⎝ ⎠ μg ⎪ 10 -------------------3 αs ≥ 0.4
Res/vrs
⎩ ( 1− αs)
vrs = 0.5[A − 0.06Res
⎧ 2/α
2
+ (0.06Res) + 0.12Res(2B − A) + A ]
2 ⎪ -s 0.01 < αs ≤ 0.1
(A.11) F1= ⎨ -------------
40 (A.22)
⎪
Where ⎩ 0.11+ 0.00051 exp (11.6αs ) αs > 0.1

Korean J. Chem. Eng.(Vol. 31, No. 7)

1160 M. Lungu et al.

⎧ 1+ 3 αs/2 + ( 135/64)αsIn(αs ) +17.89αs Radial distribution function [55]

⎪ (1− w) ----------------------------------------------------------------------------------------------
-
⎪ 1+ 0.681α −11.03α
2
+15.41α
3 2 3
⎪ s s 1+ 2.5αs + 4.5904αs + 4.515439αs
⎪ g0 = --------------------------------------------------------------------------------
- (A.36)
F2 = ⎨ αs αs ⎞ 3 0.67802
+ w 10 ------------------ -3 αs < 0.4 (A.23) 1− ⎛ ------------ -
⎪ (1− αs) ⎝ αs, max⎠
⎪
⎪ αs
⎪ 10 ------------------3- αs ≥ 0.4
JOHNSON AND JACKSON BOUNDARY
⎩ (1− αs)
CONDITIONS [34]
⎧ 0.9351αs + 0.03667 αs < 0.0953
F3 = ⎨ (A.24) 6μsαs, max ∂us, w
5
⎩ 0.0673+ 0.212αs + 0.0232/(1− αs ) αs ≥ 0.0953 us, w = − -------------------------------
- ----------- (A.37)
πφ pρsg0 3Θs ∂n
(1− αs )ρg ug − us ds
Res = ------------------------------------------- (A.25) 2 3/2
2μg κ sΘs ∂Θw 3πφ pαsus, slipg0Θs
Θw = − ----------
- ---------- + -----------------------------------------------
- (A.38)
( −10( 0.4−α s)/α s) γw ∂n 6αs, maxγw
w=e (A.26)
Where
CONSTITUTIVE EQUATIONS BASED ON THE
2 3/2
KINETIC THEORY OF GRANULAR FLOW 3(1− ew)αsρsg0Θs
γw = -------------------------------------------------
- (A.39)
4αs, max
Solid phase stress tensor
= CONSTITUTIVE EQUATIONS FOR THERMAL
τs = αsμs[∇us + (∇us) ] + αs⎛ξs − --- μs⎞ ∇ ⋅ usI
= T 2
(A.27)
⎝ 3 ⎠ ENERGY BALANCE
Gas phase stress tensor The volumetric interphase heat transfer coefficient is obtained
T 2 = by multiplying the specific interfacial area of a spherical particle
τ=s = αgμg[∇ug + (∇ug ) ] − --- αgμg(∇ ⋅ ug)I (A.28)
3 and the fluid particle heat transfer coefficient giving us the follow-
ing equation:
Solids pressure [22]
ps = αsρs[1+ 2( 1+ e)αsg0 ]Θs (A.29) 6αsh
hsg = ----------
- (A.40)
ds
Bulk solid viscosity [22]
Where h is the fluid particle heat transfer coefficient given by
4 Θ
ξs = ---αsρsdsg0(1+ e) ------s (A.30)
3 π Nu ⋅ κ g, o
h = ------------------ (A.41)
Shear viscosity of solids ds

10ρsds πΘs 2 Where Nu is the Nusselt number given by

- 1+ 4---( 1+ e)g0αs
μs, kin = ------------------------------ (A.31)
96αs(1+ e )g0 5 2 0.2 1/3
Nu = (7 −10αg + 5αg )(1+ 0.7Res Pr )
Kinetic viscosity, μs, kin [19] 2 0.7 1/3
+ (1.33 − 2.4αg +12αg)Res Pr (A.42)
10ρsds πΘs 2
- 1+ 4--- (1+ e)g0αs
μs, kin = ---------------------------- (A.32) Where
96( 1+ e )g0 5

Collisional viscosity, μs, col [19] μgcp, g

Pr = ------------- (A.43)
κg, o
4 2 Θ
μs, col = --- αs ρsdsg0( 1+ e) ------s (A.33)
5 π Phase thermal conductivities
An expression for the effective thermal bed conductivity which
Frictional viscosity μs, fr [23] consists of the gas phase conductivity and solid phase conductivity
ps sin φ was derived by Kuipers et al. [5] and has been implemented in our
μs, fr = --------------
- (A.34) model. The gas phase and solid phase conductivities are given by:
2 I2D

Where (1− 1− αg)κ g, o

κg = ------------------------------------
- (A.44)
φ is the internal angle of friction and I2D is the second invariant αg
of the deviatoric stress tensor.
Collisional dissipation of solid particle fluctuating energy [22] (βA + (1− β )K)κg, o
κs = --------------------------------------------
- (A.45)
1− αs
12(1− e )g 2 3/2
γs = -------------------------0ρsαs Θs (A.35)
ds π Where
July, 2014
 Computational fluid dynamics simulations of interphase heat transfer in a bubbling fluidized bed 1161

β =7.26×10−3 (A.48)
A −1 B A -----------
-2 ---- In ---- − B −1 − 0.5(B +1 )
2
K = ----------- ------------------ (A.46)
B ⎛1− --- B-⎞ A B 1− --- B- K is the effective thermal conductivity of a cylinder. This cylinder
1− ----
A ⎝ A⎠ A consists of one particle and the fluid phase.
κs, o β is the ratio of the particle contact area to the particle surface
A = -------
- area
kg, o
B is the factor of deformation
1− α 10/9
B =1.25⎛ ------------g-⎞ (A.47)
⎝ αg ⎠

Korean J. Chem. Eng.(Vol. 31, No. 7)

View publication stats